arxiv_searchLABSCO SUMMARY
A literature search that ends in the paper.
The chain a researcher actually walks — find the paper, follow its citations, read the PDF, check whether a dataset exists — happens in one place, without losing the thread between sources.
What it is
A research server for papers and datasets: arXiv search and download, ACL Anthology search, Semantic Scholar citations and references, Hugging Face dataset search, and web search through Exa, Brave or Tavily, plus a page crawler, LaTeX compilation and PDF reading.
What you get
- arXiv searched and a paper downloaded for reading in full
- ACL Anthology searched for computational-linguistics work
- A paper's citations and its references pulled from Semantic Scholar
- Hugging Face datasets searched by what you need them to contain
- The open web searched, through Exa or a general search tool
- A specific page fetched and read when a search result is not enough
Requirements
Python 3.12 or newer; pip install academia-mcp. It speaks stdio, SSE or streamable HTTP, defaulting to HTTP on port 5056. Web search needs a key for whichever provider you use — Exa, Brave or Tavily — and token authentication for the HTTP transports is available but off by default.
Setup effort
One command — pip3 install academia-mcp
USE CASES
01
Finding out what has been published on a problem
arXiv covers the preprints and the Anthology covers the computational-linguistics venues, so the sweep is not limited to whichever one you happen to use.
02
Following a paper's influence
Read who cited it and what it cited, which is how you find the newer work and the foundation underneath in one move.
03
Reading the paper, not the abstract
The PDF is downloaded and read, so a claim can be checked against the method section instead of the summary.
04
Finding a dataset to test an idea on
Hugging Face is searched by what the data has to contain, before you commit to building something to collect it.
05
Checking a claim against the open web
Search, then open the specific page — the difference between a citation and a link that sounded right.
21 tools
Listed in project docs. Your client may show a different list.
arxiv_downloadanthology_searchhf_datasets_searchs2_get_citationss2_get_referencesvisit_webpageweb_searchexa_web_searchbrave_web_searchtavily_web_searchget_latex_templates_listget_latex_templatecompile_latexread_pdfdownload_pdf_paperreview_pdf_paperdocument_qaextract_bitflip_infogenerate_research_proposalsscore_research_proposalsBEFORE YOU INSTALL
What to know first
Tools you get
21 tools named in the project's documentation.
Requires
- python3 on PATH
- Python 3.12+
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
pip3 install academia-mcpMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"academia": {
"command": "python3",
"args": [
"-m",
"academia_mcp",
"--transport",
"stdio"
]
}
}
}
